X-MimeOLE: Produced By Microsoft Exchange V6.5
Received: by onstor-exch02.onstor.net 
	id <01C79814.0377618E@onstor-exch02.onstor.net>; Wed, 16 May 2007 16:43:32 -0700
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Content-class: urn:content-classes:message
Subject: RE: Memo: 1.3.3 upgrade issue
Date: Wed, 16 May 2007 16:43:32 -0700
Message-ID: <BB375AF679D4A34E9CA8DFA650E2B04E03C0BD8E@onstor-exch02.onstor.net>
In-Reply-To: <BB375AF679D4A34E9CA8DFA650E2B04E03C0BD44@onstor-exch02.onstor.net>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: Memo: 1.3.3 upgrade issue
Thread-Index: AceYAJDtWO6hnyamQrCeGT5k8lsXXQADpBoAAAEHUnA=
From: "Eric Barrett" <eric.barrett@onstor.com>
To: "Jay Michlin" <jay.michlin@onstor.com>,
	"Paul Hammer" <paul.hammer@onstor.com>
Cc: "Andy Sharp" <andy.sharp@onstor.com>,
	"Tim Gardner" <tim.gardner@onstor.com>,
	"Jerry Lopatin" <jerry.lopatin@onstor.com>,
	"Caeli Collins" <caeli.collins@onstor.com>,
	"Ed Kwan" <ed.kwan@onstor.com>

I'd like to add that any field upgrade with 1.3.3's "system upgrade -s"
has "system compare -s" run manually after it.   This is documented in
the upgrade guide and the release notes, and is known by everybody in
Support, both front- and backline.  If the "system compare" fails, the
upgrade is attempted a second time, and if that fails, we usually go to
some extreme measure like shipping a flash card we've burned manually.

We're pretty much done with 1.3.3.X except a few hardcore customers who
refuse to move to 2.2 or have very very slow upgrade cycles (such as
Time Inc.).  Those who are on it are on it and probably won't move to a
new patch release.  At this stage of the game, the effort required to
backport the fixes, even if moderate, may not outweigh the cost.  I'm
not the one to make that call, of course, but if I were, I don't think
I'd bother.




-----Original Message-----
From: Jay Michlin=20
Sent: Wednesday, May 16, 2007 4:17 PM
To: Paul Hammer
Cc: Eric Barrett; Andy Sharp; Tim Gardner; Jerry Lopatin; Caeli Collins
Subject: FW: Memo: 1.3.3 upgrade issue

Paul,

We did indeed decide that these fixes to upgrade would go into 1.3.3 in
addition to Lambo. They cured stop-ship-class install file corruptions.
I asked Eric about this a couple of weeks ago, and he was aware that the
fixes did not go into 1.3.3. He also recalled that we committed to them
there, specifically at CS request. He didn't know why the fixes never
went in, and I don't either.

I think we need to do two things:

1. Get these fixes into 1.3.3.x and into the field.

2. Develop a process for following up and auditing items committed for
patches. Right now I'm specifically concerned about the "upgrade to new
flash format" we had planned for a "2.2.3" release. The Delorean core
team requested this, and if that decision is to be reversed, it ought to
require some process and discussion among those affected.

jay



-----Original Message-----
From: Andy Sharp=20
Sent: Wednesday, May 16, 2007 2:24 PM
To: Jay Michlin
Cc: Jerry Lopatin; Tim Gardner; Caeli Collins
Subject: Memo: 1.3.3 upgrade issue

It's come to my attention that the upgrade fixes have never been
integrated into the 1.3.3 branch, but that we continue to release
"patches" or point-point releases from that branch.

I want to make it clear that there can be no good reason not to
integrate these fixes.  The risk of file corruption when running the
upgrade code in that branch is near 100%.  Even in the absence of
exacerbating factors, like low memory condition,  the risk is still very
high.  There's no reason to be taking this risk with our customers or
our support resources.  The support implications of this problem are
legion: random failures in any part of the system, often without any
kind of telltale symptom.  I believe it's extremely likely we are taking
a large support hit because of this, possibly without even knowing it.

Three months ago Development, Support and QA met and decided that these
fixes were going into 1.3.3.10, but it didn't happen for some reason.
Therefore I don't think we need to gather a core team meeting or
anything, we just need to cut the red tape.

I don't know enough about the internal processes of this company to know
who, but I strongly suggest that the right person(s) be directed to
integrate those fixes into the 1.3.3 line.  It certainly won't hurt, and
it might help a lot.

Thank you,

Andrew Sharp
